AI Bots Achieve 100% Success Rate in Solving Image-Based CAPTCHAs, New Research Reveals

AI Bots Achieve 100% Success Rate in Solving Image-Based CAPTCHAs, New Research Reveals


A recent study from ETH Zurich University has revealed that advanced AI systems are now capable of bypassing CAPTCHAs—the security measures designed to distinguish between human users and bots online. The findings highlight the increasing sophistication of AI, raising concerns about the future effectiveness of CAPTCHA systems in safeguarding websites from automated abuse.

Plesner, Vontobel, and Wattenhofer's research investigates the application of sophisticated machine learning methods to answer Google's reCAPTCHAv2 system's captchas.


"We evaluate the effectiveness of automated systems in solving captchas by utilizing advanced YOLO models for image segmentation and classification. Our main result is that we can solve 100% of the captchas, while previous work only solved 68-71%. Furthermore, our findings suggest that there is no significant difference in the number of challenges humans and bots must solve to pass the captchas in reCAPTCHAv2. This implies that current AI technologies can exploit advanced image-based captchas. We also look under the hood of reCAPTCHAv2, and find evidence that reCAPTCHAv2 is heavily based on cookie and browser history data when evaluating whether a user is human or not."

According to earlier studies, CAPTCHAs are less successful than other options; Oedipus's average success rate was 63.5%. The ETH Zurich study calls for more investigation into alternate human verification techniques or the development of captcha systems that can adjust to the complexity of artificial intelligence.

Honeypots, time-based form submissions, enhanced or redesigned CAPTCHA systems, Multi-Factor Authentication (MFA), biometric authentication, and bot protection software are some alternatives to CAPTCHAs. Cloudflare Turnstile, DataDome, hCaptcha, and Friendly Captcha are commercial examples. However, given how quickly advanced AI is developing, it is unclear how long the current alternatives will last and whether they will remain successful.